Method _parse

pattern/web/__init__.py:3334–3364

(self, data, format=None)

        return self.content

    def _parse(self, data, format=None):

        # The output will be ugly: it may be useful for mining but probably not for displaying.
        # You can also try PDF(data, format="html") to preserve some layout information.
        from pdf.pdfinterp import PDFResourceManager, process_pdf
        from pdf.converter import TextConverter, HTMLConverter
        from pdf.layout    import LAParams
        s = ""
        m = PDFResourceManager()
        try:
            # Given data is a PDF file path.
            data = os.path.exists(data) and open(data) or StringIO.StringIO(data)
        except TypeError:
            # Given data is a PDF string.
            data = StringIO.StringIO(data)
        try:
            stream = StringIO.StringIO()
            parser = format=="html" and HTMLConverter or TextConverter
            parser = parser(m, stream, codec="utf-8", laparams=LAParams())
            process_pdf(m, parser, data, set(), maxpages=0, password="")
        except Exception, e:
            raise PDFParseError, str(e)
        s = stream.getvalue()
        s = decode_utf8(s)
        s = s.strip()
        s = re.sub(r"([a-z])\-\n", "\\1", s)        # Join hyphenated words.
        s = s.replace("\n\n", "<!-- paragraph -->") # Preserve paragraph spacing.
        s = s.replace("\n", " ")
        s = s.replace("<!-- paragraph -->", "\n\n")
        s = collapse_spaces(s)
        return s